About QEMU install Debian/Ubuntu on KS/OVH from RescueMode

aRNoLD

I tried this on a KS machine, and here is my process with the references from a few posts/instructions, but weird that operations are partially sucessfully. Don't know if anybody could offer their insights to help.

Below are links that I refered to:

• https://lowendtalk.com/discussion/160256/help-with-windows-server-2019-kimsufi-qemu
• https://lowendtalk.com/discussion/160282/windows-kimsufi-qemu-fresh-install-probably-will-work-for-others-too/
• https://nicolas.busseneau.fr/en/blog/2021/07/ovh-soyoustart-install-iso-image
• 
• https://rinaldojonathan.com/en/cara-buat-virtual-kvm-di-server-ovh-kimsufi-soyoustart/

These links helped when I installed AlmaLinux v8.5 (http://ash.mirrors.clouvider.net/almalinux/8.5/isos/x86_64/AlmaLinux-8.5-x86_64-minimal.iso), after which I could SSH onto the machine and do yum update or ping 1.1.1.1, or curl https://lowendtalk.com/.

The thing confused me is that when I tried Ubuntu 20 or Debian 10, the installation were all successful, installation via Text mode or Graph mode, but the server couldn't be accessed via SSH. Judging from Kimsufi's control panel, the server is running (there is a green OK button beside Server Status).

I once guessed that the installation was not successful was due to the install mode as I did the AlmaLinux in Graph mode while in the first round of Debian/Ubuntu in Text mode, but failure came the same even if I switched to Graph mode or even changed installation media, from mini.iso to netinst.iso. These iso files are all obtained from the distro's official website so there shouldn't be wrong with the iso files.

So there are the few lines of scripts in the Rescue mode

apt update && apt install -y qemu-kvm

wget -O /root/install.iso http://ftp.debian.org/debian/dists/buster/main/installer-amd64/current/images/netboot/mini.iso

qemu-system-x86_64 \
  -enable-kvm \
  -cdrom /root/install.iso \
  -hda /dev/sda \
  -m 2G \
  -net nic -net user \
  -vga std \
  -vnc localhost:0 \
  -usbdevice tablet \
  -boot d

After the VNC is initiated in the QEMU, I setup a SSH tunnel and VNC to localhost:5900 to start installation. Region/location/Repo URL/Time zone, all direct to US/America, HD set to use the whole disk/partition. Network hardware works OK, hard disk(2T) works too, or AlmaLinux would have failed.

So, where is the problem?

Lutung

run Qemu again without -cdrom /root/install.iso \

or copy this

 qemu-system-x86_64 \
   -enable-kvm \
   -hda /dev/sda \
   -m 2G \
   -net nic -net user \
   -vga std \
   -vnc localhost:0 \
   -usbdevice tablet \
   -boot d

Access again with VNC with your root password.

edit network interface "ens3" or whatever the name there installed
change it with "eno1"

reboot your server with disk.

I have tested this with proxmox, debian, ubuntu and almalinux.

my appologies if my words not good. english not my mother language

jmgcaguicla

In a non-FDE setup (no initrd fuckery), the most common thing that goes wrong when installing in qemu is the NIC interface name.

Figure out the "predictable" name of the interface and use that instead of the one qemu gives you. Failing that, you can add net.ifnames=0 biosdevname=0 to the kernel params so you get the old-style ethX names, should be useful if you only have 1 NIC.

aRNoLD

Thank you, both @Lutung and @jmgcaguicla . With your suggestions, explanations and detailed replies, I finally got the point of install failure. Now after some time invested to torture Kimsufi's netboot module and power gadget, Debian/Ubuntu splashed their screen.

I converted to Debian about two years ago and thought CentOS, even seven as bit outdated. I didn't expect such a thing that Alma sustained my stupidity. The key point after several hours turns out to be the naming convention.